A 15-person marketing agency was growing, and its weekly client reports could not keep up. These are the questions an agency owner would ask about a project like this, answered with what happened here.
What was wrong with the old reports?
Staff built every report by hand. They pulled numbers from Google Ads, Meta and HubSpot, then put them together for each client. That took about 25 staff hours a week. Reports often went out days late, or not at all. The numbers also changed depending on who built the report.
Clients noticed. They did not see steady updates, and some grew frustrated and left. Leadership had no single view of how campaigns were doing across accounts.
What does a report look like now?
Each day, n8n pulls the campaign numbers from Google Ads, Meta Ads and HubSpot into one Google Sheet. OpenAI GPT-4 reads the numbers and writes a summary of what matters. Each week, the system builds a branded Google Slides deck for every client, with the client’s logo, and emails it every Monday morning with that written commentary.
How does the team know when an account needs help?
An alert goes to the account manager when a number moves outside its set range, for example a drop in click-through rate or a budget overspend. A Notion dashboard shows every client’s campaigns in one place, marked red, yellow or green. Slack messages flag campaign tasks that are overdue.
How long did it take to set up?
Eight weeks. The first 2 mapped the data sources and set the measures and slide layouts for each type of client. The next 4 built the data pulls, the reports and the alerts. The last 2 trained the account managers and added each client’s branding. A check runs each week to make sure the data connections still work.
What changed?
| Measure | Before | After |
|---|---|---|
| Staff hours on reports | 25 a week | 2 a week |
| Client churn | 15% | 5% |
| Client NPS score | 63 | 82 |
Three clients who were at risk of leaving renewed. Every report now uses the same data, so the numbers match. Leadership sees all campaigns live. The agency puts the yearly return at $40,000 in staff time and kept clients.
What does the client say?
“Reports now send themselves.. accurate, on time, and branded. Clients actually comment on how professional it looks. This automation alone freed a full workday for every account manager.”
- Managing Director, Marketing Agency
What is next?
Each month, the team checks the AI commentary for accuracy. In month 3, invoices started going out attached to the reports. The plan is to add SEO and analytics dashboards next.